Supported Linters

We currently support the following linters:

LanguageLinters
Allcodespell, cspell, gitleaks, git-diff-check
Ansibleansible-lint
Bashshellcheck, shfmt
Bazel, Starlarkbuildifier
C, C++, Protobufclang-format, clang-tidy, include-what-you-use
Cloudformationcfnlint
CSS, SCSSstylelint
Cuecue-fmt
Dockerhadolint
Dotenvdotenv-linter
GitHubactionlint
Gogofmt, golangci-lint, semgrep, goimports
HAMLhaml-lint
Javasemgrep
JavaScript, TypeScript, YAML, JSONeslint, prettier, semgrep
Kotlindetekt1, detekt-explicit1, detekt-gradle1, ktlint
Markdownmarkdownlint
Protobufbuf-breaking, buf-lint
Pythonautopep8, bandit, black, flake8, isort, mypy, pylint, semgrep, yapf
Rubybrakeman, rubocop, rufo, semgrep, standardrb
Rustclippy, rustfmt
Scalascalafmt
SQLsql-formatter, sqlfluff
SVGsvgo
Terraformterraform (validate and fmt), tflint2
TOMLtaplo
YAMLprettier, semgrep, yamllint

  1. Support for Detekt is under active development; see its docs for more details.
  2. Module inspection, deep Checking, and setting variables are not currently supported.

Caching

Caching is currently enabled for about half the linters/formatters. Since Trunk needs to know all the inputs to a linter for a file/directory in order to cache the results, we don't yet cache every linter, but we are building out the functionality to do so. You can see which linters are currently configured to be cached by running trunk print-config and seeing which linter configurations have cache_results: true set. That's also how you can enable/disable caching for any custom linters you integrate.

Linter-specific Configuration

Most linters provide some mechanism to tweak their configuration, e.g. .eslintrc or Cargo.toml. trunk is aware of all the ways individual tools are configured and supports them. This means that all the linters you've already configured will continue to work exactly the same, just now supercharged by trunk.

Check out our open source configs repo for our always up to date collection of sane linter configurations.

Ansible-lint

Ansible-lint must be configured with a trigger. See the trigger rules documentation for more information.

In case your ansible setup is not contained within a single folder you would list all files and directories belonging to your ansible setup.

Clang-format

We currently only support versions 12.0.0, 12.0.1, and 13.0.0 ([email protected] in trunk.yaml)

By default trunk uses clang-format to additionally format .proto files. However, for this to work, you need to have enabled clang-format to do so in your .clang-format config file. You can do that by adding the following to the end of your .clang-format:

---
Language: Proto

For example, you might have this for your entire .clang-format file:

BasedOnStyle: Google
ColumnLimit: 100
---
Language: Cpp
DerivePointerAlignment: false
---
Language: Proto

Clang-tidy

We currently only support versions 12.0.0, 12.0.1, and 13.0.0 ([email protected] in trunk.yaml), and we only support using clang-tidy from Bazel and CMake projects.

In order to only see issues in your own code, not from library header files your code includes, add this to your .clang-tidy file:

HeaderFilterRegex: \./.+

You may have to build your project first if you depend on any generated header files.

Linter Failures

If a file you're linting does not compile, clang-tidy may fail to process it. In trunk, this will show up as a "Linter Failure". The output you'll see will look like a compilation error. This can also happen if the pre-reqs to running clang-tidy haven't been met (see below).

Using Bazel

By default trunk will query bazel for compile commands used to run clang-tidy. This requires no configuration.

Trunk will build needed compilation pre-requisites before invoking clang-tidy on each file (e.g. generated protobuf headers).

Finding the bazel binary

Trunk will search for the bazel binary in two ways.

  1. Paths relative to the workspace root.
  2. Binaries in any of the directories in the PATH environment variable.

First trunk will search all workspace root relative paths and then all system directories. If you override anything in lint.bazel.paths then we only search the paths you specify. By default the configuration is as follows.

lint:
  bazel:
    paths:
      workspace:
        - tools/bazel
        - bazelisk
      system:
        - bazel
        - bazelisk

Using compile_commands.json generated by CMake

Trunk supports using the compile_commands.json file generated by CMake. This file is generated by specifying the -DCMAKE_EXPORT_COMPILE_COMMANDS=ON flag when running CMake. This will generate a compile_commands.json file in your build directory. You must then symlink this file to the root of your repository for trunk and other tooling to find it. For example:

ln -s /path/to/project/build/compile_commands.json /path/to/project/compile_commands.json

You must also configure trunk to use compile_commands.json by adding the following to your trunk.yaml:

lint:
  compile_commands: json

<other tool> claims I have clang-tidy issues, but not trunk- what gives?

Trunk runs clang-tidy with a compile commands database so that we can guarantee clang-tidy produces the correct diagnostics about your code. Other tools, such as clangd, may use best-effort heuristics to guess a compile command for a given clang-tidy input file (for example, see this discussion) and consequently produce incorrect clang-tidy findings because they guessed the compile command wrong.

Clippy

Clippy is distributed with rust itself, so specify your rust version for your clippy version (for example [email protected]).

Detekt

Detekt is usually invoked through gradle, which allows specifying additional configuration in build.gradle. We do not yet automatically parse your Gradle scripts to infer your detekt configuration; instead, what we do is this:

  • detekt invokes detekt-cli with the
    --build-upon-default-config flag (this appears to be
    more common than the
    alternative)

  • detekt-explicit invokes detekt-cli without the
    --build-upon-default-config flag

You will also need to provide a valid detekt config as .detekt.yaml (an empty .detekt.yaml is valid, if you don't want to configure detekt); if you already have a detekt config, then you can symlink it like so:

ln -s path/to/existing/detekt-config.yml .detekt-config.yaml

To use ./gradlew detekt to invoke Detekt, you can add [email protected] to your enabled list. Note that since you're running Detekt via Gradle, you should also add the paths to your Detekt configurations to direct_configs, e.g.

direct_configs: ["lib/detekt.yaml"]

Eslint

Most eslint users use a number of plugins, custom parsers, etc. Trunk has the ability to turn sandboxing and caching on or off for each linter, and we've turned it off for eslint so it can use your repo's installed packages for eslint plugins and other required eslint packages. Trunk does control the eslint version itself, but otherwise eslint looks for all plugins, configs, etc based on the path of source file its linting. This all means you do need to have npm/yarn installd in your repo as a prerequisite before running eslint via trunk.

We recommend you disable all prettier rules in your eslint config and let trunk run prettier automatically on your files. It's much nicer to just autoformat a file than to see a lint error for every missing space.

You can easily do this by adding the eslint-config-prettier package and in your eslint config's extends section adding prettier as the last element. For example, your extends list might look like:

extends:
  # Order matters, later configs purposefully override settings from earlier configs
  - eslint:recommended
  - airbnb
  - plugin:@typescript-eslint/recommended
  - plugin:import/recommended
  - plugin:import/typescript
  - plugin:node/recommended
  - plugin:mocha/recommended
  - plugin:react/recommended
  - prettier # this actually turns OFF all prettier rules running via eslint

Python linters (flake8, pylint, black, etc)

Trunk uses hermetic runtime versions, which you can override if needed. If you're using a newer version of python that our default (3.10.3 at the time of writing) you can override it in trunk.yaml via:

runtimes:
  enabled:
    - [email protected]

As always, you can view the defaults and configuration of everything trunk runs via trunk print-config.

Flake8

Flake8 has a plugin architecture where if you install a plugin, it gets used. You can enable Flake8 plugins via:

enabled:
  - [email protected]:
      packages:
        - [email protected]

flake8-bugbear is probably the most popular flake8 plugin, we recommend it!

Prettier

By default trunk uses prettier to autoformat many languages/config formats, including markdown. To line wrap within markdown, you need to set the following in your prettier config (.prettierrc.yaml or similar):

proseWrap: always

You may also want to configure printWidth: to your liking.

Gitleaks

Gitleaks v7 only works with go 1.16, not go 1.18 while Gitleaks v8 works with 1.18. We recommend using v8, but if you specifically need to use v7 you can override the go runtime version like so:

runtimes:
  enabled:
    - [email protected]

Again, this is not recommended. Just use Gitleaks v8 or later with go 1.18 or later.

Golangci-lint

Make sure your go version in go.mod matches Trunk's go runtime version. At the time you writing, Trunk's default go runtime version is 1.18.3. You can find out what it is via trunk print-config, and look for the runtime section, and you can override the default version in your trunk.yaml via:

runtimes:
  enabled:
    - [email protected]

Pylint

You may specify additional pylint plugins in your .pylintrc, using the line load-plugins=...

if you want to run the plugin pylint-django as part of your setup, you would add the line

load-plugins=pylint_django to your .pylintrc, but you also need to tell trunk to install the package:

- [email protected]:
    packages:
      - [email protected]
      ...

Markdownlint

Older versions of markdownlint had a bug where it printed plaintext output even when run with --json. We rely on JSON output so we can parse and ingest the results from markdownlint. The package we use for markdownlint is actually markdownlint-cli; >= 0.29.0 is verified to work.

Rustfmt

We currently use the version of rustfmt packaged with rust, so for rustfmt version, specify your rust version (for example [email protected]).

If you have edition in your cargo.toml, rustfmt also needs the same information in .rustfmt.toml in your repo root. For example, your .rustfmt.toml might contain:

edition = "2021"

Sqlfluff

sqlfluff is only configured as a linter because its formatting capabilities are very subpar compared to sql-formatter.

Terraform

We currently support terraform validate and terraform fmt, but only fmt is enabled by default when you add terraform to your enabled list in trunk.yaml. To enable validate, add this to your trunk.yaml:

lint:
  linters:
    - name: terraform
      commands:
        - name: validate
          enabled: true
  enabled:
    - ...

Note: you must run terraform init before running trunk check with terraform validate enabled (both locally, or on CI).


Did this page help you?